Knowledge graphs are large-scale semantic networks that considerably impact knowledge representation. Mining hidden knowledge from existing data, including triplet knowledge reasoning, is a primary objective of knowledge graphs. With the development of Neural Network (NN) and Deep Learning (DL), the interpretability of triplet knowledge reasoning gradually decreases; furthermore, what machines learn is not actual reasoning but digital reasoning shortcuts. To solve this problem, more background knowledge needs to be introduced into knowledge graphs: causal graphs can offer valuable causal logic knowledge for reasoning; temporal quadruples can provide essential temporal distribution details; and commonsense graphs can furnish pertinent commonsense understanding to support reasoning. In recent years, many scholars have incorporated additional background knowledge into knowledge graphs to construct more complex reasoning mechanisms. This paper reviews the basic concepts and definitions of knowledge reasoning and the reasoning methods used for knowledge graphs. Specifically, we dissect the reasoning methods into four categories: triplet reasoning, causal inference, temporal inference, and commonsense reasoning. Finally, we discuss the remaining challenges and research opportunities related to knowledge graph reasoning.

产品设计是产品开发过程中的创造性阶段,是产品开发过程中最重要的环节之一.产品设计技术、方法和工具已经成为决定设计能力与水平的关键要素.近年来,产品设计方法朝着基于知识和数据驱动两个方向快速发展,推动产品设计的自动化和智能化水平提升.本文论述了一种知识与数据融合驱动的智能设计技术,该技术结合了基于知识和数据驱动两种方法的优势,给出了知识与数据融合驱动的智能设计技术框架,论述了设计知识分类与量化表征、设计数据获取与试验设计、知识与数据融合模型、基于智能优化算法的方案生成等关键技术.上述技术可应用于航天器气动力预测、航天控制力矩陀螺试验设计、水下发射装置流场重构等工程问题中,验证了知识与数据融合技术解决工程问题的潜力及优势.

随着系统工程建模技术的进步,复杂装备生命周期所涉及的建模工具多样,模型数据异构为模型集成带来了重大技术挑战.因此,系统工程社区提出了语义建模规范并将其用于解决建模工具之间的模型数据集成难题,提升各工具建模语言的数据互用性.提出一个新的语义式系统工程建模语言KARMA,采用双语义机制支持复杂装备研发过程中的架构建模,并提出了该语言的技术发展路线.最后,通过工程验证案例,从定性及定量角度对语言进行技术验证,结果表明KARMA语言可支持多种复杂装备的体系及系统的架构设计.

针对以引力波探测为代表的空间科学任务和以"国网星座计划"为代表的商用卫星网络任务对推进器的特殊需求,本文提出了一种基于数字孪生的推进器优化设计方法.该方法首先建立由机理模型模块和测试数据集模块组成的数字孪生体.机理模型模块依据推进器的物理过程建立模型,对难以测量的数据进行仿真模拟;测试数据集模块通过实验对推进器进行测试,依靠测试数据建立可测参数的数学模型.将数字孪生体与实验进行对比,通过对比结果反馈调节机理模块从而不断提高孪生体的准确性,最终为优化设计提供依据.结果表明:(1)该方法能够构建微波离子推进器的数字孪生体;(2)该数字孪生体的预测结果存在一定差异,通过分析发现该差异与机理模型的精细度以及测试数据集的数据量有关.

为推动航天器以物理试验为主导的模式向以物理试验与虚拟试验深度融合的模式转型,针对航天器试验过程中存在的周期长、成本高与其他制造环节海量数据共享困难等问题,提出基于数字孪生和多智能体的智能试验体系架构设计方法.该方法首先建立包含物理空间、数字空间和数学空间的智能试验任务分析范式,基于数字孪生和多智能体的智能试验体系功能建模框架;然后针对智能试验全要素、全过程、全业务的应用背景,设计试验资源、试验过程和试验服务3类智能体的结构模型,并对试验运行机制和技术流程进行优化.以某航天器系统级正样热试验为例搭建了智能试验管控平台,对该方法的可行性和有效性进行验证.结果表明,智能试验管控平台的应用,使得产品试验设计与实施总周期缩短20％、产品试验与工装设计精度提高15％.

针对现有复杂武器装备多工序制造系统(MSMS)模型描述不全面、分析结果不准确的问题,在充分分析MSMS特点的基础上,提出质量传递综合模型(QTCM),建立了多工序质量传递过程的严格数学表达.在QTCM的框架范围内,分别借助心理学和统计学知识,对以往难以量化的操作者能力和制造设备状态进行量化.采用制造过程历史数据构建影响因素和质量输出之间的映射关系,建立了质量的传递过程.通过分析各影响因素对产品质量偏差的贡献量,确定影响多工序制造质量的主要误差源,进而为产品质量的提升提供指导.以某型号导弹发动机转子系统为例进行了验证,结果表明,QTCM可以方便地借助原材料质量或工艺质量计算获得产品的最终质量分布状况,分析结果与工厂测量结果一致.

A method to determine the best contact state of precision assembly based on data registration is proposed for the effects of form error and its distribution on the assembly precision and the unknown con tact state after assembly.The distribution character of geometric form error is analyzed.A method to determine the contact point is proposed using data registration technology.And the best contact state is solved based on particle swarm optimization (PSO).On this basis,the small displacement torsor (SDT) is used to express the form error of part surface,and the contact error is calculated to predict assembly accuracy based on contact point.The proposed method is verified by a case study.The results show that it is of great significance to predict the assembly precision by using the data registration method to determine the assembly contact state.Meanwhile,the distribution of form error has also an important influence on the assembly precision,which cannot be ignored in the research on precision assembly.

In order to develop a quantitative evaluation index model for the essential characters of RMS (reconfigurable manufacturing system) and to optimally select the configuration plans with their outstanding advantage and disadvantage comparison,an evaluation method was proposed based on PROMETHEE (preference ranking organization method for enrichment evaluation).Analyzing the reconfiguration of the reconfigurable machine tool and manufacturing cell,a basic evaluation index and the quantitative model was designed to take the key characters of RMS,scalability,convertibility,diagnosability,modularity,integrability and customization.AHP (analytic hierarchy process) was used to assign the weight for these indexes.In the evaluation process,PROMETHEE I was applied firstly to analyze the advantage and disadvantage for each alternative plan.Secondly,PROMETHEE Ⅱ was adopted to analyze the comprehensive advantage.Finally,the sequence of alternative plans was decided based on the analyzed results.A case study from a shop floor of an institute was presented to validate the effectiveness and practicability.

In the process of multistage precision assembly,geometric error of parts is one of the important factors influencing on the assembly precision,and the method to determine the geometric error's influence and the bottleneck assembly are the primary problems.In this paper,the pose change of the parts was analyzed to concern the mating surface geometric error caused by surface topography in two parts assembly,and to determine the mating error of the single process.Based on the multi-body system theory,an error transformation model of multistage assembly process was established;and then,a sensitivity analysis model of the assembly precision was built with matrix differential method to identify the main geometric errors for the assembly system.The method provides theoretical basis for precision assembly accuracy analysis and control.Finally,an example was presented to validate the established model.

The traditional form error evaluation methods,based on the principles of a minimum tolerance,are inadequate to quantitatively describe the vital influence of the distribution of the surface form error on the assembly precision.In this paper,a method was proposed to determine the potential evaluation parameters of surface form error for assembly precision.Firstly,nonGaussian surface construction was used to simulate machining surface,and wavelet analysis with the selected base function was used to obtain the surface form error.The input parameters of non Gauss simulation were determined by analyzing the test parameters of the "real machined surface",and the candidate parameters characterizing the surface form error distribution were selected.Then the virtual assembly of the filtered surface with ideal smooth surface was conducted,hence the ideal surface spatial location was calculated.The relationship between the surface parameters and the second surface spatial location was established by correlation analysis to determine the form error evaluation parameters with a certain criteria.Finally a case was used to verify the feasibility proposed method.The method can provide a scientific basis for optimizing the assembly process to improve assembly precision.

为了开拓设计思路、产生创新产品方案,将设计知识有效融入产品的概念设计过程中,提出一种融合设计过程与设计知识的产品概念设计方法.定义了基于产品—功能—结构的过程模型和知识模型,利用逐层映射行为、回溯映射行为、检索行为和存储行为支持设计过程和设计知识的融合.在产品域,通过基于组合权重的双层灰色关联分析法得到待改进产品的相似产品集;在功能域,使用功能相似度算法和功能操作方法得到新产品功能架构;在结构域,利用形态学矩阵生成多种概念方案,提出基于粗数的逼近理想解排序法,并定义了综合约束指数,对方案进行定性和定量评价、得到最优方案.利用所提方法获得了一种新型美容补水仪的概念方案,从而验证了该方法的可行性.

针对目前本体中概念相似度求解方法孤立考虑概念信息量或概念关系的问题,提出基于场论的概念相似度求解方法.该方法首先明确概念的语义关系是由概念包含的信息决定且隐性的概念关系通过显性的概念关系推出,并将概念包含的信息划分为共性信息与特性信息两类.结合场论的原理和方法,将本体映射为与物理场具有相同特性的语义场,将概念的共性信息和特性信息分别转化为语义引力元和语义距离,采用引力场模型计算概念的相似度.结合坦克装甲车辆和舰船领域知识检索需求,建立基于本体的语义检索系统,并通过与五种典型的相似度求解方法进行对比分析,验证了该方法的正确性和有效性.

For the problem that reconfigurable manufacturing systems (RMS) have to consider the efficiency and flexibility at the same time, a method for the formation of part family, which considers bypassing moves and idle machines, is presented. First, the longest common subsequence (LCS) among different process of parts is identified. Then, on basis of LCS, a shortest composite supersequence (SCS) is formatted through the combination of LCS and the rest operation under the consideration of bypassing moves and idle machines. Based on the linear relationship between parts similarity and LCS as well as SCS, the similarity coefficient algorithm is designed. Finally, the developed similarity coefficient has been compared with the exiting best similarity coefficients available in the existing literature and the accuracy and efficiency has been verified.

为提高夹具设计过程中的设计知识重用率和设计效率,提出一种基于本体和知识组件的夹具结构智能设计方法。该方法通过构建夹具元件、工件以及夹具设计实例的本体模型,实现了对夹具设计知识的表示。通过引入知识组件技术并建立本体间的语义映射关系,构建针对夹具选择、尺寸驱动和建模装配的三种知识组件及知识处理框架,形成夹具结构智能设计模式,完成了对夹具结构设计知识的组织、推送和嵌入。结合基于规则和实例的推理方法,实现了对夹具的智能选择,通过开发相应的夹具结构设计系统,实现了基于本体参数化属性的夹具尺寸自驱动和基于本体装配特征的夹具元件快速组装。以某飞机零件为例,对所提方法的可行性和有效性进行了验证。

为了提高基于实例推理方法在枪械方案设计领域的检索精度和设计知识重用程度,提出针对变型设计过程的基于实例推理的设计方案获取和方案修改方法.针对枪械设计过程,建立包括知识库、推理驱动引擎、实例检索模块和规则修改模块的实例推理流程;根据3类不同的指标需求形式设计了对应的相似度计算方法,提高检索精度;通过规则知识的定义和约束实现对设计方案进行定量化、适应性修改,获取最终的方案.开发了基于实例推理的枪械方案快速设计原型系统,并以枪管为例验证了该检索算法的精度,以及所提方法的有效性和实用性.

针对目前网络化制造环境下设备分组和检索算法效率不高的现状,提出了基于设备加工特征向量的分组方法来提高算法的效率.首先,按照设备所能完成的加工特征,从特征的形状、尺寸以及加工精度等三个维度构建了制造设备的加工特征描述向量;然后基于制造设备的加工特征描述向量,通过设计了初始分组中心选择策略和采用非遍历分组组数优化技术,改进了扩展模糊C-均值算法,构建了基于加工特征向量的设备分组模糊聚类算法.该算法可高效、精确地计算出设备最优分组组数.最后,以50台制造设备的分组为例验证了该算法的有效性.

To deal with setup planning problems in computer aided process planning,a novel setup planning method based on clustering analysis and tolerance reasoning was proposed orienting to machining features.By analyzing manufacturing process of part,the machining units were defined and a mathematical model for setup planning was established.Initial solution of setup planning was randomly generated,and dissimilarity matrix of machining units was calculated.The minimum group was formed by iterative clustering analysis.By tolerance reasoning,sequence constraint of machining units was acquired and setup planning sequences within group and between groups were generated.The optimal setup planning was put forward.Based on CATIA design platform,setup planning prototype system was developed and the proposed algorithm was verified by an example.

To deal with setup planning in computer aided process planning, a novel setup planning method based on memetic algorithm is proposed. By analyzing geometric characteristics of the part, machining features and units are determined and representation of setup planning is established. The initialize population of setup planning is configured by candidate tool approach direction, machines and cutter for each machining unit. Setup planning is searched in the global scope by partial mapped crossover and insertion mutation. Based on sequence constraints between units, binary tree sort algorithm is adopted to transform from infeasible solution to feasible solution. The sum of the processing methods similarity between machining units is taken as fitness function, setup planning of high fitness value can be acquired in local search by crossover operation based on fitness rate and mutation operation of non-sequential constraint machining units. After the evolution of populations, optimal setup planning solution is generated. Setup planning process of typical part is illustrated to prove the feasibility of the proposed model.

To speed up the pro duct design efficiency, product designers would like to utilize the previous design knowledge. This requires a systematic and structured way for design knowledge representation and reuse. A new method which is named knowledge component to make design knowledge reused conveniently is presented in this paper. Knowledge component is a virtual reuse model which has specific function. The internal structure and working principle of knowledge component are discussed; meanwhile the involved design knowledge was analyzed in detail. The design process of barrel chamber was taken as an example to illustrate the executing process of knowledge component. Testing result shows that this method is feasible in terms of increasing design efficiency and helps the designers reuse the existing knowledge rapidly.